© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 65960
    Prev
    Next
    INDEX
    Explanations

    punching and boxing

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron fires on present‐participle descriptions of physical violence—most strongly on “punching” (and its token parts).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     għ
    -0.76
    Marseille
    -0.75
    ycznego
    -0.75
     SOU
    -0.75
    Athlete
    -0.70
    睞
    -0.70
     Sunglasses
    -0.69
     Sunscreen
    -0.69
    澡
    -0.69
    Parametro
    -0.68
    POSITIVE LOGITS
     punching
    2.78
    Punch
    1.93
    punch
    1.92
     punch
    1.90
     Punch
    1.88
     bag
    1.61
     punched
    1.46
     bags
    1.36
     PUNCH
    1.34
     punches
    1.31
    Activations Density 0.023%

    No Known Activations